A Nature editorial summarizing 31 papers argues that pure data-driven artificial intelligence (AI) is insufficient for clinical adoption and that precision medicine requires integrating structural domain knowledge before model training. The editorial identifies three necessary transitions: anchoring algorithms in biological laws for mechanistic interpretability; designing collaborative, human-centered clinical interfaces; and scaling equitable, population-level interventions. The collection spans video analysis, signal analysis, digital health integration, and population health, arguing that medical AI’s future lies in minimalistic, interpretable tools combining domain knowledge with state-of-the-art algorithms to improve patient outcomes and equity. Video-based diagnostics are described as transitioning from controlled laboratory settings to portable, accessible tools capable of specialist-level accuracy in everyday environments; signal analysis is pivoting toward lightweight devices performing complex tasks in natural environments; wearable and biomarker integration now capture dynamic health complexity through multivariate models; and population health applications are demonstrating that AI’s ultimate test lies in whether it can deliver equitable benefit at scale. Examples include smartphone-based graph convolutional networks assessing Parkinsonian gait asymmetry at home, VideoPose3D extracting accurate 3D joint angles by removing noise via clustering, open-source models transforming raw video into interpretable Autism Spectrum Disorder metrics, and smartphone systems combining video analysis with large language models for ophthalmologic evaluation. Deep learning also enables real-time ultrasound lesion detection, noninvasive blood count estimation from ocular videos, and cardiovascular risk prediction from echocardiography.